From 78f50ab8b4bd57cd67d015fa2608bc86b116dfba Mon Sep 17 00:00:00 2001 From: =?UTF-8?q?Anton=20Luka=20=C5=A0ijanec?= Date: Sun, 19 Feb 2023 20:13:13 +0100 Subject: 4chan script fix --- skripti/4chan-thread-dl.py | 12 +++++++++--- 1 file changed, 9 insertions(+), 3 deletions(-) diff --git a/skripti/4chan-thread-dl.py b/skripti/4chan-thread-dl.py index 7a71df2..f1e2717 100755 --- a/skripti/4chan-thread-dl.py +++ b/skripti/4chan-thread-dl.py @@ -6,6 +6,12 @@ if len(argv) != 3: raise Exception("prvi arg je url do threada, drugi arg je končnica, recimo .webm") s = BeautifulSoup(get(argv[1]).content, "html.parser") for a in s.find_all("a"): - if a.attrs["href"].endswith(argv[2]) and "title" in a.attrs and "/" not in a.attrs["title"]: - print(a.attrs["title"]) - open(a.attrs["title"], "wb").write(get("http:" + a.attrs["href"]).content) + if a.attrs["href"].endswith(argv[2]): + fn = None + if "title" in a.attrs and "/" not in a.attrs["title"]: + fn = a.attrs["title"] + elif "/" not in a.getText() and ("class" not in a.attrs or "fileThumb" not in a.attrs["class"]): + fn = a.getText() + if fn is not None: + print(fn) + open(fn, "wb").write(get("http:" + a.attrs["href"]).content) -- cgit v1.2.3